AWS Certified AI PractitionerResponsible AIMedium
A credit scoring AI model is found to consistently give lower scores to individuals residing in certain low-income neighborhoods, regardless of their individual financial history. This happens because the model implicitly learns correlations between zip codes (which are proxies for income) and creditworthiness from historical biased data. This phenomenon is best described as:
- AOverfitting
- BConcept Drift
- CAlgorithmic Bias
- DData Leakage
Show answer & explanationAnswer & explanation
Correct answer: C. Algorithmic Bias
Algorithmic bias occurs when an AI model's design, training data, or deployment leads to systematic and unfair discrimination against certain groups. In this case, the model learns a biased correlation from data, resulting in unfair outcomes based on neighborhood.
Why the other options are wrong
- A. Overfitting is when a model learns the training data too well, performing poorly on new data.
- B. Concept drift is when the relationship between input and output data changes over time.
- D. Data leakage is when information from outside the training dataset is used to create the model.
Algorithmic Bias
Algorithmic bias refers to systematic and repeatable errors in a computer system that create unfair outcomes, such as favoring or disfavoring particular groups of people.
- Can stem from biased training data, flawed algorithm design, or deployment context.
- Leads to discriminatory or inequitable results.
- Requires careful detection and mitigation strategies.
Memory trick: Algorithms can be 'blind' to unfairness if not carefully designed.